Co-built the original Telegram file mirroring bot that spawned a widely-forked ecosystem. Adoption was large enough that mass Heroku deployments contributed to Heroku killing their free tier.
Backend & Systems Engineer
I build systems that eliminate manual work. If I do something twice, I automate it the third time.
Go · Python · Concurrent Systems · Open Source · Ontario, Canada
I'm a backend engineer who has been building things since before anyone was paying me to. I co-authored python-aria-mirror-bot — a Telegram automation tool that got forked hundreds of times and adopted at a scale that contributed to Heroku deprecating their free tier. When its C++ SDK bindings kept crashing, I learned Go, wrote CGO bindings from scratch, and wrapped them in a REST microservice. The senior contributor to that SDK starred the repo.
That's the pattern in everything I build: find the manual process or the broken thing, and fix it properly. I'm looking for a backend or systems engineering role where I can keep doing exactly that.
Co-built the original Telegram file mirroring bot that spawned a widely-forked ecosystem. Adoption was large enough that mass Heroku deployments contributed to Heroku killing their free tier.
MegaSDK's SWIG Python bindings kept crashing across the ecosystem. Designed CGO-based C++ bindings from scratch in Go, exposed as a REST server. Eliminated the crash surface entirely. Starred by a senior MegaSDK contributor. Adopted by other projects.
Full Go rewrite of the mirror bot. Embedded a GoJa JavaScript runtime to build an extensible plugin system — scripts register via URL regex and resolve direct download links transparently before the transfer pipeline runs. Same architectural pattern as Webpack loaders or Caddy modules.
High-performance concurrent CLI for Google Drive. Recursive downloads, resume support, OAuth2, service account rotation, MD5 integrity verification, real-time progress.
Telegram bot for managing Heroku deployments without touching the dashboard. Restart dynos, stream logs, manage config vars — all from Telegram.
A five-phase operational analytics system to decide whether a brand-new regional airport (HWR, opened 2 weeks prior) was reliable enough as the first leg of an international connection to Toronto. Ingested Air India API + FlightAware ADS-B feeds, built a historical reliability database (completion rate, average delay, cancellation frequency), tracked aircraft tail-number rotations to model delay propagation, added PrevFlight risk analysis to explain why a flight might be delayed before it happens, all exposed through a Telegram bot with live ASCII flight progress. Answering one question with data instead of assumption.
Virtualized homelab on Proxmox. Custom VPN-router VM with Linux Netfilter and iptables for traffic routing and firewall rules. Previously patched and compiled a custom Android kernel (Linux 3.x) for a Samsung device. This portfolio is hosted on it.